Update t_test.py

[mohazahran]

for high d values the call to power_ttest will invoke the c++ boost lib that will throw an exception (below) that will not be caught here.
We can temporeary circumvent this behavior by saturating the d value at 4 (for statistical_power == 0.9 and pvalue == 0.1) and return
a pre-computed required sample size for this value.

[balikasg]

Minor Questions:

• Should we be using " instead of ' as in the docstring?
• There seems to be an extra indent in the details compared to the '''
• Please start with capital (For)
• Typos: temporeary

Main questions:

• How is 2.24 calculated?
• What's the plan to actually fix the issue and remove this temporary fix? Is there one?
• power==0.9 and pvalue==0.1 are very specifix. Should we have something more generic e.g., power>=0.9 and pvalue<=0.1?

Thanks for catching this and trying to fix it here. Is there already an open issue on power_ttest?